AI-Driven Test Authoring — quiz

You ask Claude (with Playwright MCP) to 'write a test that lets a user buy a t-shirt.' The first draft uses fragile CSS selectors and hard-coded waits. What is the most effective follow-up prompt?

• Repeat the same prompt verbatim
• Constrain the output: 'use Playwright's `getByRole`, `getByLabel`, and `getByText` locators, no `waitForTimeout`, and structure the test in arrange/act/assert with explicit network or visibility waits' — and provide one example test from the existing suite as context
• Ask the model to make the selectors longer
• Switch the prompt to a different language